Brazilian law presents what might be called the foundational paradox of posthumous personality protection: the same legal system that declares personality extinguished at death simultaneously legislates as though something of it persists. Article 6 of the Civil Code of 2002 establishes that the existence of a natural person ends with death. Article 11 reinforces this by providing that personality rights are non-transferable and irrevocable. Yet articles 12 and 20 extend legal protection to the image, voice, and honor of deceased persons, granting spouses, lineal relatives, and collateral relatives up to the fourth degree the standing to demand that threats or injuries to those rights cease, and to claim damages.

The tension is not merely textual. It reflects a deeper jurisprudential difficulty: personality rights were designed for living subjects who can exercise, waive, and defend them. When the subject disappears, the architecture falters. The Roman maxim mors omnia solvit, death resolves all things, captures the classical position, but Brazilian law has progressively deviated from it, recognizing, as [20] argues, that personality rights project themselves beyond the life of their holder. The attack on the honor of the dead does not affect the deceased, obviously, but it produces effects in the social environment, and the law aims at maximum protection for the attributes essential to the human condition. The problem is compounded by the temptation to resolve this tension through the property framing: treating the deceased’s digital information as “assets” to be governed by succession law presupposes property status without justifying it. As [3] demonstrate, the use of the term “digital assets” in policy and scholarly discussions is itself an epistemological choice that forecloses the inquiry it pretends to answer—the language of “my data” and “my account” mimics property discourse, while the underlying object is, in Floridi’s terms, closer to “my body” or “my feelings” than to “my car”.

The result is a legal structure that [19] describes as principally defensive and repressive: it prohibits abuse but does not regulate use. This distinction has enormous practical consequences in the context of posthumous digital avatars. The Civil Code provides a remedy when someone violates the posthumous image of the deceased through an offensive publication or unauthorized commercial use, but it provides no affirmative framework governing the conditions under which digital reconstruction is permissible in the first place. The law responds to harm; it does not shape the market that produces it.

A critical clarification is necessary at this point, one that Brazilian law has not always made with sufficient precision. The family members authorized by arts. 12 and 20 of the Civil Code to protect the deceased’s personality rights are not, in the technical sense, the holders of those rights. They act in their own name, in defense of a right of their own, which is the right to preserve the memory, dignity, and moral integrity of the person who is gone ([19]). The rights themselves, being non-transferable, do not pass by succession. What passes is only the standing to defend them.

This distinction between guardian and proprietor has significant implications for the digital avatar market. If heirs are guardians rather than owners, then the commercial logic that treats posthumous identity as an asset to be managed, licensed, and monetized by the estate rests on a shaky doctrinal foundation in Brazilian law. The system, as [20] observes with characteristic directness, does not convert privacy, image, and honor into things that can be transmitted by inheritance. Any exploitation of the economic potential of a deceased person’s identity requires specific legal authorization, and that authorization is precisely what current Brazilian law does not provide.

The Ninth Civil Law Conference of the Federal Justice Council (CJF) offered a partial response to this gap through Enunciado 687, recognizing that the digital patrimony may form part of the estate of a deceased holder, subject to legitimate succession. But the Enunciado was careful to note that strictly personal digital data, private messages, personal diaries, intimate albums, tend to be considered non-transferable, preserving rights of privacy and dignity ([19]). The challenge is that digital avatars do not fit neatly into either category: they are constructed from personal data, yet designed for external presentation and often commercial deployment.

The Brazilian General Data Protection Law (LGPD, Law 13.709/2018) offers what might appear to be an additional layer of protection for posthumous identity. In practice, it offers none. The LGPD defines personal data as information relating to an identified or identifiable natural person, and limits its protections to living persons, as confirmed by the National Data Protection Authority ([1]) in Technical Note 3/2023.

Responding to a consultation from the Federal Highway Police regarding the creation of an online memorial for deceased employees, the ANPD concluded with notable bluntness that the LGPD is inapplicable to the data of deceased persons. The authority grounded this conclusion in art. 6 of the Civil Code: since the existence of the natural person ends at death, so does the framework designed to protect the fundamental rights of development of living personality ([19]). The protection of post-mortem personality rights, the ANPD held, is not covered by the LGPD, since there is no longer any development of personality to protect.

This conclusion has been criticized for its excessive simplicity. As [19] notes, it ignores real conflicts that emerge in the posthumous use of personal data such as image, voice, and digital archive, leaving those situations entirely outside the safeguards of the data protection framework. Nothing in the ANPD’s interpretation would prevent third parties from creating deepfakes using the image and voice of deceased persons, reconstructing their appearance or speech without any legal basis or control, since such uses fall outside the scope of the LGPD. This structural withdrawal is consistent with the design of the GDPR, whose recital 27 explicitly excludes deceased persons from its scope while leaving Member States free to extend protection—a choice that several European jurisdictions have made, though in limited and often inconsistent ways ([3]). The Brazilian legislature’s silence is therefore not exceptional, but it is particularly consequential given the scale of the market operating in its shadow.

The regulatory gap is, in this respect, not a minor lacuna to be filled by interpretive creativity. It is a structural absence that coincides precisely with the most commercially significant zone of the posthumous identity market. The LGPD’s withdrawal at the moment of death leaves the field open to market actors who are investing precisely in the data of the dead.

In the absence of a specific legal framework, the testament emerges as the most effective available instrument for ensuring that the deceased person’s own wishes prevail regarding posthumous uses of their image and voice. Brazilian Civil Code art. 1.899 establishes that testamentary clauses susceptible to different interpretations shall be resolved in favor of the interpretation that best ensures compliance with the testator’s will. Arts. 1.857 and 1.846 establish the general framework of testamentary freedom, constrained only by the protection of the compulsory share (legítima) reserved for necessary heirs.

[20] observes that the legal order does not admit a full waiver of personality rights, but does recognize the validity of limited, purpose-specific arrangements. By analogy, a testator may legitimately specify in their will the conditions and duration of any posthumous uses of their image and voice, including outright prohibition or, conversely, specific authorization for identified purposes. As [19] notes, citing Sanches, personality rights do not cease to exist with death, they merely end their natural development, and therefore need to be respected and maintained in accordance with the will of their holder, just as they would have been during life.

The cases of Robin Williams and Madonna illustrate the stakes. Williams reportedly structured his estate to include a twenty-five-year restriction on the commercial use of his image and likeness, an effective private solution to a problem that public law had not yet addressed ([19]). Madonna has publicly stated her opposition to holographic tours after her death, concerned about image distortion and the potential for exploitation by others ([19]). Both cases exemplify what the Civil Code reform commission has begun to call ante-mortem autonomy: the use of instruments created during life to govern identity after death.

The limitation of the testamentary instrument is obvious: it requires foresight, legal awareness, and access to legal counsel that are not equally distributed in society. A person who dies without having thought to address posthumous image uses in their will leaves their family and, ultimately, the courts with no authoritative guidance. The law cannot rely on individual initiative to fill a regulatory gap of this magnitude.

Ultimately, while the testament offers an important mechanism for expressing ante-mortem autonomy, it cannot serve as a complete regulatory solution. Its effectiveness depends on individual awareness and access to legal planning, conditions that are unevenly distributed and that many people do not meet. As technologies capable of recreating voices, images, and identities continue to expand, relying solely on private testamentary arrangements leaves significant uncertainty for families, courts, and society. This limitation reinforces the need for clearer legal frameworks capable of safeguarding posthumous personality rights and ensuring that the dignity and intentions of individuals are respected even in the absence of explicit testamentary guidance.

The posthumous digital avatar phenomenon does not merely expose a gap in Brazilian law. It exposes something more fundamental: a structural crisis in the doctrine of personality rights itself, whose cardinal attributes, namely non-transferability (intransferibilidade), irrevocability (irrenunciábilidade), extrapatrimoniality, and inescapability (imprescritibilidade), were designed for living persons and prove increasingly inadequate when confronted with technologies that effectively dissolve the boundary between identity and commodity. This crisis is not purely theoretical: it has direct practical consequences for the regulation of an industry that has built its business model precisely on the attributes that personality rights doctrine declares non-transferable.

The attribute of intransferibilidade, established by art. 11 of the Civil Code, is the node around which the crisis is most acute. In classic doctrine, the inalienability of personality rights serves a dual protective function: it prevents the rights-holder from alienating their own identity under conditions of duress or inequality, and it prevents third parties from acquiring rights over another person’s attributes of personhood. Both functions were designed with the living person in mind. Applied to the deceased, inalienability produces a paradox: the rights cannot be transferred, yet someone must exercise them, and whoever exercises them will, in practice, have the power to license, or refuse to license, the commercial exploitation of the deceased’s identity. The doctrine insists on calling this “guardianship” rather than “ownership”, but the functional difference becomes difficult to sustain when the guardian’s decisions determine who profits from the deceased’s reconstructed voice, image, and interactive persona.

The digital avatar market intensifies this crisis along several dimensions. First, it transforms inalienable identity attributes into raw inputs for a production process: the voice, the image, the verbal patterns, the expressive style of the deceased are captured, processed, and recombined by machine learning systems to produce new content that the deceased never generated. What the platform acquires is not formally a “right” to the deceased’s personality, but a data corpus from which synthetic identity can be generated at scale. Inalienability doctrine, calibrated to prevent the transfer of rights, has no ready response to a market that operates not by transferring rights but by extracting data and generating new outputs. This is precisely the mechanism that [3] identify as the defining characteristic of the personal data category among digital remains: unlike intangible items or copyright works, personal data do not comprise property during people’s lives, and hence there is no reason to convert them into property upon death—a conclusion that applies with even greater force to the synthetic outputs generated from that data. The legal object that needs protecting, namely the integrity of posthumous identity, is not the same as the legal object that personality rights doctrine was designed to protect.

Second, the market exploits what might be called the “consent gap” that inalienability doctrine produces. Because personality rights cannot be alienated, their commercial exploitation, even where permitted, must always take the form of a license or authorization, never an assignment. But the persons who provide that authorization in the posthumous context (heirs and legal representatives) do so in their capacity as guardians of a right they do not hold, not as owners of a right they may freely dispose of. This creates a structural tension: the persons whose consent is legally required are not the persons whose identity is at stake, and their interests may diverge from the deceased’s in ways that the law currently has no mechanism to detect or correct. [20] identified this problem with characteristic precision, noting that the legitimation of heirs as the sole guardians of posthumous personality interests risks converting protection of the dead into exploitation by the living.

Third, and perhaps most fundamentally, the digital avatar market challenges the extrapatrimonial character of personality rights by making the economic value of posthumous identity both enormous and specifically quantifiable. Classic doctrine treats the patrimonial dimension of personality rights (the economic value of image and voice, for example) as an incidental byproduct of rights whose primary nature is existential, not commercial. But when a platform offers to generate an interactive simulation of a deceased celebrity for subscription fees, the economic dimension is not incidental: it is the entire purpose of the enterprise. The Brazilian civilian tradition has, as noted above, been gradually adapting to this reality through jurisprudential evolution (as in the STJ’s Garrincha cases) and through the hybrid framework proposed by the Civil Code reform commission. But adaptation has been piecemeal and reactive, driven by litigation rather than by principled doctrinal revision.

The crisis of inalienability in this context produces four specific regulatory challenges. First, there is the challenge of identification: determining who bears the burden of demonstrating that a proposed posthumous digital use is within the limits of legitimate guardianship, rather than an exercise of de facto ownership. Second, there is the challenge of calibration: developing legal standards robust enough to distinguish permissible licensing of posthumous identity (consistent with the deceased’s presumed will and dignity) from impermissible commercialization (which exploits the deceased’s identity for purposes they would have rejected). Third, there is the challenge of enforcement: building mechanisms capable of detecting, and sanctioning, uses that nominally comply with the formal consent requirements but violate the substantive integrity constraints that non-transferability is meant to protect. And fourth, there is the challenge of adaptation: determining whether the concept of inalienability itself needs to be reformulated, by distinguishing between the existential core of personality (absolutely inalienable, permanently protected against commercial capture) and the expressive surface of identity (capable of being licensed, within strict substantive limits, for uses consistent with the deceased’s own mode of self-presentation), or whether its current formulation can be made adequate through interpretive development alone. The Civil Code reform commission’s proposals suggest the former: a reformulation, rather than a mere reinterpretation, is needed. Whether the legislative process will deliver one is the central question facing Brazilian law in this domain.

French law treats the right to one’s image as an extension of the right to private life ([2]), grounded in art. 9 of the French Civil Code, which establishes that everyone has the right to respect for their private life. The doctrinal position of the Cour de cassation has been consistent: the right to one’s image, being a personality right, is extinguished at death and is not transmissible to heirs. A 2009 decision of the First Civil Chamber made this explicit, holding that the court of appeal had correctly found that family members had no standing to make claims based on the commercial exploitation of the artist’s image, since that right, as an attribute of personality, ended with the death of its holder ([19]).

What survives in the French system is not a right held by heirs, but a narrowly defined standing to seek compensation where the publication of a deceased person’s image causes them a proven personal harm, typically a moral injury resulting from an offense to the deceased’s memory ([19]). This is a considerably more limited protection than a property right: the family cannot authorize commercial uses of the deceased’s image, because there is no right to authorize; they can only react to abuses that injure them personally. This structure reflects a foundational design choice: the French system prioritizes the non-commercial, moral dimension of personality, treating image rights as too intimate to be marketized, even posthumously ([14]; Cairn.info/Légicom, 1995).

In 2016, [7] moved to partially address the digital dimension of this framework through the Loi pour une République Numérique (Law 2016-1321), which introduced the concept of post-mortem digital directives. Any person may now specify instructions regarding the conservation, deletion, and transmission of their personal data after death, including their image. These directives may be general, covering all personal data, or specific to identified services. In the absence of directives, heirs retain subsidiary rights to access data necessary for estate settlement, close online accounts, or oppose the continuation of data processing relating to the deceased ([19]). The law also requires service providers to inform users about what will happen to their data after death and to allow them to designate a trusted third party to carry out their instructions. Despite this legislative step, the Loi pour une République Numérique remains focused on data governance rather than identity rights, and does not alter the fundamental position of the Cour de cassation that personality rights, including the right to one’s image, are extinguished at death ([8]).

The implications for posthumous digital avatars are significant. The creation of a deepfake or avatar of a deceased person without prior authorization runs into the principles already outlined: absent consent by the person or the family, any public dissemination may be challenged by heirs if it causes them suffering or indignation. More specifically, art. 226-8 of the French Penal Code, as amended by Law 2024-449 of May 21, 2024, criminalizes the bringing to public knowledge of any montage made with the words or image of a person without their consent, including content generated by algorithmic processing that represents a person’s image or words, unless it is manifest that the content has been algorithmically generated or is expressly stated to be so. Penalties rise to two years’ imprisonment and 45,000 euros for offenses committed via online communication services ([19]). French criminal law thus functions as a guarantor of posthumous memory: absent authorization, the law prevents any deceptive reproduction of the deceased person, regardless of whether the family formally authorizes it.

The French model’s coherence comes at a price. By refusing to recognize any patrimonial dimension of posthumous personality rights, it leaves families without legal tools to assert any economic stake in the commercial exploitation of their relative’s identity, even where that exploitation is substantial. In the digital avatar context, this means that a company could in principle use the image of a deceased French person for commercial purposes, provided it obtained the necessary authorizations and respected the applicable disclosure requirements, without owing anything to the family beyond the avoidance of a moral offense.

German law has charted a different course, one that the Bundesgerichtshof (BGH, Federal Court of Justice) developed progressively through landmark decisions beginning in the late twentieth century. The doctrinal starting point is the general right of personality (allgemeines Persönlichkeitsrecht), grounded in arts. 1 and 2 of the Grundgesetz (Basic Law), which protects human dignity and the right to the free development of personality. The traditional view held that this right was strictly personal and non-transferable, with the consequence that after death only immaterial interests, honor, reputation, dignity, could be protected by family members, and only through injunctive relief, not financial compensation. This doctrinal framework, grounding personality rights in the constitutional guarantees of human dignity and freedom (Arts. 1 and 2 GG), distinguishes the German approach sharply from both the French model, which lacks a general personality right, and the Californian model, which is rooted in property rather than dignity ([17]; Max Planck Encyclopedia of European Private Law, 2012).

The 1999 Marlene Dietrich decision (BGH, 1 ZR 49/97) transformed this framework. Maria Riva, the sole heir of the actress who died in 1992, brought proceedings against the unauthorized commercial exploitation of her mother’s name and image. The BGH recognized, for the first time, that the patrimonial components of personality rights persist after death and pass to the heirs. These components, the commercial value that a person’s image and name command in the marketplace, were found to be heritable, with the consequence that the heir could seek not only injunctive relief but also financial compensation for unauthorized uses ([19]). The court specified that the heir must exercise these rights in accordance with the expressed or presumed will of the deceased. The Marlene Dietrich decision thus marked a conceptual turn in German law: by recognizing both patrimonial and non-patrimonial dimensions within a unitary framework, it avoided the dichotomy between “privacy rights” and “publicity rights” that characterizes American law, and produced a model in which commercial value and human dignity are not treated as mutually exclusive ([9]; [6]). It is significant, however, that in the parallel BGH decision concerning Facebook’s refusal to enable parental access to a deceased teenager’s account (Case III ZR 183/17, 2018), the court grounded its ruling not in personality rights but in the inheritability of the user contract itself—treating the contractual relationship as the heritable object, not the personal data it governed ([3]). The same court that extended patrimonial personality rights in Marlene Dietrich carefully avoided recognizing posthumous data protection, relocating the legal question instead to contract and succession law. The digital avatar context puts this separation under pressure: the coherence of the German hybrid model depends on maintaining it.

The Kunsturhebergesetz (KUG, Law on the Copyright of Works of Visual Art), enacted in 1907 and still in force, provides the operative framework for posthumous image protection. Its §22 establishes that the distribution or public display of a person’s portrait requires the consent of the depicted person. After the person’s death, consent of their close relatives is required for a period of ten years. After that period, the protection lapses, and in principle the image may be used freely, subject, however, to the permanent residual limit of §23(2), which prohibits any use that would injure the legitimate interests of the family of the deceased ([19]).

The German approach thus produces a distinctive hybrid: a recognized patrimonial dimension, transmissible to heirs for a defined period, but constrained by a permanent moral floor grounded in human dignity. This design addresses one of the central weaknesses of the French model, the absence of any legal mechanism through which families can share in the economic value generated by posthumous identity, without abandoning the principled commitment to treating dignity as a non-negotiable limit on commercial use.

For the regulation of digital avatars, the German model offers instructive lessons. The temporal limit of ten years provides a degree of legal certainty that more open-ended formulations do not. The criterion of the heirs’ legitimate interests provides a doctrinal anchor for challenging uses that, even if commercially authorized, disfigure the deceased’s identity or subject them to undignified portrayals. And the requirement that heirs act in accordance with the expressed or presumed will of the deceased introduces the concept of ante-mortem autonomy into the analysis, even in a system that has gone further than France in recognizing commercial transmissibility.

The state of California has developed the most commercially significant framework for posthumous identity rights in the United States. The Right of Publicity, the right to control the commercial use of one’s name, image, and voice, has been recognized in American law since the mid-twentieth century, when courts began to distinguish between the right of privacy (a personal and non-transferable interest) and the right of publicity (a property-like interest in the commercial value of one’s identity). As a property right, the right of publicity is assignable, licensable, and transmissible at death. Its duration, governance, and specific content vary significantly by state, since there is no federal statute. This fragmentation has produced significant legal uncertainty: the applicable law typically depends on the state of the deceased’s domicile at death, leading to outcomes that are difficult to predict and inconsistent across jurisdictions ([10]; [5]). The absence of a federal standard has also meant that the explosion of AI-generated digital replicas has been met with a patchwork of state responses rather than a coherent national framework. This fragmentation reflects a theoretical inconsistency at the foundation of the property model: as [3] argue, the property framing of digital remains has a differential privacy effect—while it may strengthen heirs’ position vis-à-vis platforms, it simultaneously exposes the deceased’s personal data to family members in ways the deceased may not have intended. The Californian model, with its emphasis on commercial operationalization, systematically prioritizes the former and ignores the latter.

New York Civil Rights Law §50-f (2020) is among the most sophisticated state-level instruments, particularly for its treatment of digital replicas. The statute defines a deceased performer as a person who, for gain or livelihood, was regularly engaged in acting, singing, dancing, or playing a musical instrument, and a deceased personality as any deceased natural person whose name, voice, signature, photograph, or image has commercial value at the time of death ([19]). Both categories are protected against unauthorized commercial uses. The statute fixes the duration of protection at forty years after death, after which claims for unauthorized uses are barred.

The statute’s most innovative provision addresses digital replicas directly. It defines a digital replica as a computer-generated, originally created performance by an individual in a separately created, original expressive audio or audiovisual work in which the individual did not actually perform, that is so realistic that a reasonable observer would believe it to be a performance by the individual depicted ([19]). The use of a digital replica of a deceased performer in a scripted audiovisual work or in a live musical performance requires prior consent of the rights holders. An exception is created, however, for uses accompanied by a visible disclaimer in the credits and in related advertising stating that the use was not authorized by the rights holders, an exception that [19] aptly criticizes as a loophole, since a disclosure of non-authorization reduces only the risk of fraud claims, not the underlying violation of the personality right.

Louisiana’s Allen Toussaint Legacy Act (2022), enacted to honor the late musician and renowned New Orleans composer, extends protection to a broader category of persons: it requires no showing of prior commercial exploitation of identity, protecting any individual whose identity has commercial value, regardless of fame. The statute defines identity broadly to include digital replicas, and sets a maximum duration of fifty years after death, subject to earlier termination for non-use over a continuous three-year period ([19]). Louisiana’s approach is stricter than New York’s regarding digital replicas of performers: it focuses on whether the use creates in the public mind the impression of an authentic performance, without providing the disclaimer-based escape valve that New York law permits.

California’s Civil Code §3344.1, as amended by Assembly Bill 1836 (2024), represents the most commercially significant state framework, given Hollywood’s dominance of the global entertainment industry. The statute provides seventy-year post-mortem protection for the name, voice, signature, photograph, or likeness of deceased personalities. AB 1836 added explicit provisions for digital replicas, defined as computer-generated, highly realistic electronic representations of an individual’s voice or visual likeness, incorporated in a sound recording, image, audiovisual work, or transmission in which the individual did not actually perform or appear ([19]). The unauthorized production, distribution, or making available of such replicas in expressive audiovisual works or sound recordings is prohibited without consent of the rights holders, with minimum statutory damages of USD 10,000 per occurrence ([19]).

At the federal level, the No Artificial Intelligence Fake Replicas and Unauthorized Duplications Act (NO FAKES Act), introduced in 2023 and reintroduced in revised form in 2024 and 2025, would create a federal digital replica right, applicable to both living and deceased persons, and provide a cause of action against the unauthorized production, publication, distribution, or transmission of a digital replica ([5]). The bill’s post-mortem provisions have attracted scholarly criticism, however: [18] argue that as currently drafted, the NO FAKES Act would incentivize, and in some cases force, the commercialization of the dead against their wishes and the preferences of their families, prioritizing wealth generation over the dignity and autonomy values that should anchor posthumous protection. The ongoing federal debate thus reflects the same tension between commercial and dignitarian conceptions of posthumous identity that separates the Californian model from its European counterparts.

The Californian property model offers what the other two models do not: a clear, enforceable, commercially operational framework for the posthumous exploitation of identity. Its weakness is the inverse of its strength. [21] observed that data compilations capture the bare facts of what people do without the motives, and that the person is reconstituted in databases as a digital person composed of data, a reconstruction that may fail to represent, and may distort, who people really are. Applied to the digital avatar context, this critique suggests that the right of publicity, for all its operational clarity, may be structurally indifferent to the integrity of the posthumous persona: it asks who holds the right, not whether the use respects the person. This structural indifference has been identified as a foundational weakness of the Californian model in comparative perspective: the absence of a constitutionally grounded concept of dignity means that post-mortem personality protection depends entirely on the commercial value of the deceased’s identity, leaving those without celebrity status, or those whose families decline to commercialize their memory, without effective legal recourse ([9]; [17]).

The three models converge on one firm point: commercial use of a deceased person’s image or voice requires valid authorization, from the person in life or from whoever the law designates as successor. All three also maintain some form of post-mortem protection of dignity, memory, and identity, with the possibility of preventing disrespectful dissemination and seeking redress. All three preserve exceptions linked to freedom of expression and information, journalistic, biographical, historical, critical, satirical, and incidental uses, that cannot be conflated with advertising or marketing. These convergences reflect what [8] identifies as a shared commitment, across civil law and common law traditions alike, to preserving some minimum of posthumous dignity, even where the scope and instruments of protection differ dramatically.

Where they diverge is precisely on the two dimensions that matter most for the digital avatar market: the patrimonial transmissibility of posthumous identity rights, and the duration of protection. The French model refuses patrimonial transmissibility entirely. The German model recognizes it with a temporal limit and a moral floor. The Californian model embraces it with long durations and few substantive constraints on the purposes for which it is exercised. Brazil, as [19] notes, currently aligns with the French tradition of non-transmissibility and purely moral post-mortem protection. The challenge is to determine whether that alignment remains defensible in a market environment where identity has become a renewable resource.

The Brazilian Civil Code of 2002 was drafted before digital avatars were a commercial category, and it shows. The code provides no definition of the legal nature of digital audiovisual content left by the deceased, no framework governing the conditions under which such content may be reproduced or reconstructed by AI systems, and no mechanism through which the deceased person’s own wishes about posthumous uses can be operationalized in a legally binding way beyond the general instrument of the testament. [19] characterizes this omission with precision: the current framework addresses only the repressive dimension of post-mortem protection, providing remedies against abuse but not affirmative governance of legitimate use.

The LGPD’s withdrawal from posthumous data, confirmed by the [1] in Technical Note 3/2023, compounds the problem. The ANPD acknowledged that the law was conceived to protect the fundamental rights of persons alive, and that the post-mortem protection of data subjects’ personality rights is therefore not covered ([19]). The authority noted that certain civil law mechanisms, the Civil Code’s protection of posthumous name and image under arts. 12 and 20, remain applicable, but these were designed for a world of photographs and written materials, not for large-scale AI-driven reconstruction of identity.

The legislative vacuum described in the preceding subsection did not go uncontested. Three bills submitted to the Brazilian legislature in 2023 reflect early attempts to address the regulatory gap, and the tensions between them foreshadow the doctrinal choices that would later be addressed, more systematically, by the Civil Code reform commission. Their content is worth briefly recounting, not as a definitive regulatory response, but as evidence of the depth of disagreement within the Brazilian legal system on the fundamental questions of consent and authorization.

Bill 3608/2023 (Deputy Jadyel Alencar) and Bill 3614/2023 (Deputy Benedita da Silva) staked out opposing positions on the pivotal question of heirs’ authority. The former conditioned any posthumous deepfake on prior and express ante-mortem consent, but assigned the power of authorization and the entitlement to commercial gains to heirs in the absence of such consent, a compromise that preserved the primacy of individual will while recognizing the family’s practical role. The latter proposed an outright prohibition on technology-assisted reconstruction of the deceased’s voice or image absent a contrary testamentary disposition, and crucially denied heirs the power to authorize such reconstruction at all, a position grounded in [20] observation that heir-guardianship risks converting the protection of the dead into a vehicle for the economic interests of the living. A third bill (3592/2023, Senate) adopted a middle position granting heirs decision-making authority in the absence of ante-mortem directives; it was ultimately declared moot (prejudicado) in 2024 and archived, its substantive concerns having been absorbed into broader legislative initiatives. The irresolvable divergence between these bills on the role of heirs (guardians or de facto proprietors) was precisely the tension that the Civil Code reform commission would later attempt to resolve through its dual-authorization framework, discussed in the following subsection.

In April 2024, the Commission of Jurists designated by the Brazilian Senate to propose a comprehensive revision of the Civil Code concluded its work, presenting an anteprojeto (draft reform) covering over one thousand updated articles. Among its most significant innovations was the inclusion of a dedicated chapter on Digital Law within the Civil Code, addressing digital patrimony, artificial intelligence, personal data protection, and electronic contracts ([19]).

The reform proposal classifies digital assets into three categories: strictly personal digital assets of an existential or intimate character; digital assets of intrinsic economic value; and hybrid digital assets that combine personal and economic dimensions. The third category, which expressly includes images and videos, is the most relevant for the purposes of this article. The proposal’s art. 1.791-A provides that the deceased’s digital assets of appreciable economic value form part of the estate, while §2 of the same article establishes that personality rights projecting beyond death without economic content, including privacy, intimacy, image, name, honor, and personal data, are governed by special law and the general personality rights chapter of the Code ([19]).

On the specific question of AI-generated images of living and deceased persons, the reform commission proposed a provision permitting such creation for lawful activities, subject to: prior and express informed consent of the person, or of heirs and representatives of the deceased; respect for the dignity, reputation, and legacy of the natural person represented, avoiding uses contrary to their mode of being or thinking as externalized in life; prior and express authorization of spouses, heirs, or representatives, or by testamentary provision, for commercial use; and absolute respect for mandatory rules and public order ([19]). The commission’s final report explicitly acknowledges the importance of treating posthumous digital assets as hybrids: transmissible as to patrimonial aspects with consent, but indisposable as to the existential core of identity, dignity, and memory.

This framework is promising, but its operationalization leaves significant gaps. As [19] observes, the criterion of coherence with the mode of being requires documentary evidence, writings, statements, public practices, and interpretive parameters that discourage distortions. The requirement of transparency must be accompanied by minimum governance duties for platforms and producers, including registration of consents, editing logs, and dispute and takedown procedures, lest labeling become a merely formal gesture.

The practical stakes of the reform commission’s framework are well illustrated by the CONAR adjudication of the Volkswagen/Elis Regina case, in which Brazil’s advertising self-regulatory body cleared a commercial featuring a digitally reconstructed performance by the deceased singer, finding that the heirs’ authorization, the active participation of her daughter Maria Rita in the production, and the coherence of the portrayal with the artist’s established persona were sufficient to legitimate the use ([19]). The case is instructive precisely because it confirms that consent and identity coherence function as workable criteria even under existing soft-law mechanisms, but also because it exposes the limits of that soft-law approach: CONAR’s decisions bind only its members, carry no legal force beyond the advertising context, and leave the deeper questions about heirs’ authority and applicable legal standards entirely unresolved. The reform commission’s proposed framework would, if enacted, supply the binding legal standard that the CONAR adjudication necessarily lacked, transforming what was an ad hoc regulatory judgment into a generalizable, judicially enforceable norm.

The first principle is that the expressed will of the living person regarding posthumous uses of their identity must function as both the upper limit and the primary instrument of legitimate consent. Where a person has, in life, specified their wishes regarding the creation, distribution, or commercialization of digital avatars of themselves after death, whether through a testament, a specific contractual instrument, a digital directive of the kind now recognized in French law, or any other legally valid expression of will, that specification should bind family members, platforms, and courts.

This principle has several implications. It means that legal systems should actively facilitate the creation of clear, accessible instruments for ante-mortem expression of identity preferences, and should treat the absence of such expression not as license but as a reason for caution. It means that heirs should be understood as implementing the presumed will of the deceased, not as exercising an independent proprietorial right, a distinction that both the Brazilian Civil Code tradition ([20]) and the German BGH’s Marlene Dietrich formulation recognize. And it means that technological and platform governance should be designed to make ante-mortem expression possible and visible: platforms collecting data that could be used for posthumous reconstruction should be required to provide users with the means to specify their posthumous preferences and to ensure that those preferences are respected.

The weight of this obligation, however, is not uniform across service types. For archival retrieval systems, ante-mortem autonomy is satisfied when the person consented, during their lifetime, to the collection and posthumous use of their own recordings. For synthetic generative systems, ante-mortem autonomy requires a qualitatively higher standard of consent: the person must have expressly authorized not merely the use of their existing data, but the creation of new content in their name, a form of authorization that existing legal instruments—the Brazilian testament, the French digital directive, the Californian publicity rights assignment—are generally not designed to capture with the specificity that synthetic generation demands.

[4] has argued that the majority framing of posthumous digital technologies in terms of grief management has obscured their commercial dimensions and created a false impression of benevolence around an industry that raises fundamental questions of power and consent. Ante-mortem autonomy as a regulatory principle addresses this directly: it insists that the person whose identity is being commercialized must have had the opportunity to shape the terms on which that commercialization occurs.

The importance of ante-mortem autonomy is further underscored by the philosophical debate over whether restrictions on grief-bot access may themselves constitute a form of unjust paternalism. [16] argues that policies aimed at limiting bereaved individuals’ access to grief-bots, whether in the form of complete bans, therapist-supervised access, or technological guardrails such as time limits and programmed “uncanny moments”, constitute what he terms affective paternalism: a non-consensual intervention in a subject’s emotional lifeworld, undertaken in the name of improving their affective situation. Drawing on relational autonomy theory and the feminist tradition ([22]; [15]), Podosky contends that such restrictions disproportionately harm marginalised subjects, who already operate within systems that deny recognition of their capacity for emotional self-regulation. From a regulatory standpoint, this argument cautions against overly restrictive regimes that privilege the judgment of institutions over the ante-mortem choices of the individual, reinforcing the case for facilitating genuine autonomous expression of consent prior to death rather than delegating post-mortem decisions to platforms or heirs.

The second principle is that the content generated by digital avatars must be coherent with the identity that the person constructed and expressed during their lifetime. This is the criterion that the legislative bills preceding the Civil Code reform, the reform commission’s own proposals, and the CONAR adjudication of the Volkswagen/Elis Regina case have all, in different ways, converged upon. It represents a translation into regulatory language of the doctrinal insight that the protection of posthumous personality is not merely a protection of family interests, but a protection of the person themselves, of the legacy they built, the values they held, and the way they wished to be understood.

As a regulatory principle, integrity of the posthumous persona has both substantive and procedural dimensions. Substantively, it prohibits the production of content that attributes to the deceased words, actions, opinions, or associations that are clearly incompatible with their established identity, the Bruce Lee whisky advertisement being a paradigm case ([19]). It also prohibits uses that degrade, ridicule, or sexualize the deceased in ways that violate the basic dignity that the legal system continues to protect after death. Procedurally, it requires that those authorizing posthumous digital uses demonstrate that they have assessed compatibility with the deceased’s established identity, and that they bear responsibility for breaches of that compatibility.

The practical application of this integrity standard differs fundamentally depending on the service type. For an archival retrieval system, coherence verification is in principle tractable: the content to be presented is drawn from a fixed corpus of the deceased’s own words, and the question is whether the selection and contextualization of that corpus is faithful to the person’s identity. For a synthetic generative system, the verification task is categorically harder: the content to be assessed has never existed before, is produced by a model that may extrapolate well beyond the training data, and cannot be compared against any statement the person actually made. Integrity of the posthumous persona, as a legal standard, demands correspondingly stricter scrutiny (and a higher threshold of authorization) for synthetic generation than for archival retrieval.

[12] identify authenticity as one of the core principles of their ethics consensus framework for digital duplicates: digital duplicates should not be used in ways that are deceptive or that misrepresent the views or identity of the person depicted. Applied to the posthumous context, this principle takes on additional weight: the deceased cannot speak for themselves, cannot correct misrepresentations, and cannot withdraw consent. The asymmetry between the platform’s capacity to generate content and the deceased’s incapacity to respond makes authenticity and coherence not merely ethical desiderata but legal requirements.

The third principle follows from the analysis of the comparative models and the Brazilian civilian tradition. Family members authorized to protect the posthumous personality rights of the deceased are guardians of those rights, not proprietors of them. This distinction must be operationalized in legal frameworks that specify what guardians may and may not do.

What guardians may do includes: defending the deceased’s identity against unauthorized uses by third parties; authorizing uses that are consistent with the deceased’s presumed preferences and with the integrity of the posthumous persona; declining to authorize uses that, in their reasonable judgment, would be incompatible with the deceased’s values or would harm the deceased’s memory; and, where the deceased’s identity has recognized commercial value, participating in the economic fruits of authorized uses in a manner that reflects their role as stewards rather than owners.

What guardians may not do includes: authorizing uses that violate the moral core of posthumous personality rights, regardless of the economic returns they would generate; acting as though they hold a property right that they may freely alienate, license to anyone, or exploit in any manner; and displacing the deceased’s own expressed preferences in favor of their own economic interests. The critique articulated by [20], that the current legitimation of heirs alone may serve primarily to protect the heirs’ financial interests rather than the deceased person’s dignity, must be incorporated into any framework that extends authority to heirs. The theoretical basis for this constraint is well established: as [3] demonstrate through their analysis of will theory and interest theory, the law can at most protect the reasonable expectations of the living regarding their post-mortem condition—not the interests of a person who no longer exists as a legal subject. The implication for guardianship is direct: heirs act legitimately only insofar as they implement those ante-mortem expectations, not insofar as they exercise an independent authority derived from their position as successors.

The guardian/proprietor distinction also maps differently onto the two service categories. In archival retrieval systems, the guardian’s role is primarily curatorial: they control access to a corpus that the deceased themselves created. The risk of overreach is real but bounded by the content of the corpus itself. In synthetic generative systems, the guardian’s role is effectively constitutive: they authorize the creation of new content that will be attributed to the deceased. This is a qualitatively different power, and one that the civilian concept of guardianship, which was built around the idea of protecting what the deceased left behind, was not designed to confer. Legal frameworks should therefore impose stricter substantive and procedural constraints on heir authorization of synthetic generation than on heir authorization of archival access.

[4] observation that heirs authorized to permit posthumous digital reconstruction may themselves be in a state of acute grief, and therefore structurally vulnerable to pressure by platforms seeking to commercialize the deceased’s identity, adds a consumer protection dimension to this principle. The legal framework must ensure that the authorization process is genuinely free, informed, and untainted by the exploitation of grief, requirements that existing contract law and consumer protection frameworks may not be sufficient to guarantee.

The fourth principle is that companies offering posthumous digital reconstruction services must bear affirmative governance obligations proportionate to the risks they create. The existing practice, visible in the terms of service of platforms such as HereAfter AI, StoryFile Life, and DeepBrain AI, is that platforms claim extensive rights over user-generated content, grant users only limited and revocable licenses, prohibit any commercial use or transfer, and impose contractual conditions designed to manage the platform’s legal exposure without necessarily protecting the interests of users, their families, or the deceased ([19]). This is a governance structure designed for the platform’s benefit, not for the governance of a technology with significant human rights implications. [3] identify this structural feature as a general characteristic of the digital remains market: in the absence of legislation, the power to determine the fate of digital remains lies mostly with the service providers, who draft contracts to maximize operational flexibility and minimize exposure, while social norms that once governed posthumous access to personal items are rendered irrelevant by the intermediated, contractual architecture of online services.

Platform accountability, as a regulatory principle, requires at minimum: transparency about how data is collected, processed, and used to construct posthumous avatars; mechanisms enabling users and their families to specify, revise, and enforce posthumous data preferences; governance obligations requiring platforms to document consents, maintain records of generated content, and respond to takedown requests from authorized guardians; and liability for uses that violate the ante-mortem preferences of the deceased or the authorization decisions of their legal guardians.

Platform accountability obligations should also be calibrated to service type. Archival retrieval platforms bear accountability for the accuracy and contextual fidelity of the corpus they manage, and for the security of access controls. Synthetic generative platforms bear the additional and more demanding obligation of accounting for the content they produce: they must document the basis on which each generated response was deemed compatible with the deceased’s established identity, maintain records of generated content sufficient to enable post hoc review, and bear strict liability for generated content that violates the integrity of the posthumous persona, regardless of whether a family member nominally authorized the use. The distinction matters because the harm profile of the two categories differs: archival retrieval can distort through selection and decontextualization, while synthetic generation can distort through invention, and invention without accountability is the deeper threat to posthumous dignity.

The Harm Mitigation and Proportionality Principle proposed by [12], that the creation and use of digital duplicates is permissible only when serious harms are unlikely, or where foreseeable risks are proportionate to justifiable value and accompanied by adequate mitigation measures, provides a useful ethical anchor for platform accountability. Translated into legal terms, this principle implies a duty of care for platforms: they must assess the risk profile of each use, implement safeguards proportionate to those risks, and be answerable for foreseeable harms that their services enable.

The ethics literature on re-creation services provides concrete guidance on what platform accountability should entail in practice. [11], analyzing multiple speculative scenarios of deadbot deployment, identify four clusters of design obligations for service providers: i) the development of sensitive procedures for “retiring” deadbots in a dignified way, including automatic inactivation protocols after extended disuse, analogous to existing inactive account policies; ii) the implementation of meaningful transparency, understood not merely as disclosure that an AI is involved, but as clear communication of the psychological risks associated with prolonged interaction, including the risk of dependency; iii) the imposition of age restrictions, given the particular vulnerability of children to forming attachment relationships with AI systems representing deceased parents; and iv) adherence to a principle of mutual consent, requiring that both the data donor and the service interactant explicitly agree to participation in a re-creation project before the service commences or delivers content. The authors emphasize that this last principle is especially important because current platform designs allow data donors to designate interaction targets without the latter’s knowledge or consent, a design choice that can impose unwanted grief-processing frameworks on bereaved individuals and deny them agency over how they mourn. These design obligations translate directly into the legal accountability framework proposed here: platforms should be required by regulation to implement them, and liability should attach where failure to do so causes foreseeable harm.

The four principles outlined above are substantive; they specify what values a regulatory framework must protect. But they also require institutional translation: mechanisms through which those values are operationalized, monitored, and enforced. The CONAR adjudication of the Volkswagen/Elis Regina case illustrates both the potential and the limits of soft-law mechanisms: the advertising self-regulatory body was able to apply consent and identity coherence as workable criteria, but its decisions are non-binding, sector-specific, and unequipped to address the full range of posthumous digital uses beyond the advertising context.

The most structurally significant institutional response to this governance deficit may yet come from within the Civil Code reform process itself. The Anteprojeto de Reforma do Código Civil, finalized by the Commission of Jurists designated by the Brazilian Senate in April 2024, includes a dedicated Livro do Direito Digital that addresses, with unprecedented specificity, the legal regime of digital avatars and posthumous digital replicas. The proposal directly engages the terminological and conceptual landscape mapped at the outset of this article: whether the entities in question are understood as “avatares digitais”, “réplicas póstumas”, or “duplicatas digitais”, the reform commission proceeds from the recognition that these are sui generis legal objects, not reducible to either ordinary intellectual property or to traditional personality rights, and requiring a framework calibrated to their specific characteristics.

The reform proposal introduces a double-authorization structure that is worth examining in detail, because it departs in important respects from both the legislative bills discussed in the preceding subsection and from the current civilian doctrine. The proposed provision contemplates that both the person themselves, through an ante-mortem directive or testamentary disposition, and their heirs and legal representatives, after death, may validly authorize the construction and deployment of a digital avatar. This dual authorization model (which we may call the “conjunctive consent” framework) differs significantly from Bill 3614/2023, which would prohibit heirs from authorizing posthumous digital reconstruction entirely, and from Bill 3608/2023, which treats ante-mortem consent as a condition precedent that, in its absence, triggers derivative heir authorization. The reform commission’s approach instead treats the two sources of consent as complementary rather than hierarchical: ante-mortem authorization by the data donor is the preferred and most legally secure basis; authorization by heirs or representatives is permissible in its absence, but is bounded by the integrity requirements discussed below and by the express prohibition of uses contrary to the deceased’s established mode of being or thinking as externalized during life.

Critically, the Livro do Direito Digital reconnects this authorization framework to the doctrinal vocabulary of personality rights that the reform preserves in the general part of the Code. The draft provision establishing the conditions for lawful AI-generated images and voices of deceased persons, subject to requiring prior and express informed consent of the data donor, or of heirs and representatives; respect for dignity, reputation, and legacy; prior and express authorization for commercial uses; and absolute respect for mandatory rules and public order, operates as a lex specialis that qualifies, without displacing, the general protection of posthumous personality attributes under arts. 12 and 20 of the Code. In the terminological register adopted by this article, the reform treats posthumous digital avatars as hybrid objects: their construction and deployment engage both the existential dimension of personality (governed by the non-transferable, dignity-based rules of the general chapter) and the patrimonial dimension of digital assets (governed by the succession rules of the Livro do Direito Digital and the proposed art. 1.791-A). This bifurcation is not merely taxonomic: it has direct legal consequences for the question of who may authorize what uses, on what conditions, and subject to what limits.

The reform commission’s dual-authorization model also has significant implications for the governance relationship between public institutional mechanisms and private ordering. Rather than channeling posthumous digital use through a specific administrative body, the Livro do Direito Digital relies on a substantive legal standard, namely coherence with the established identity of the deceased, that must be demonstrated by whoever seeks to authorize the use. This approach distributes the verification function across multiple actors: notaries and testamentary executors, where ante-mortem directives exist; heirs and their legal representatives, in cases of post-mortem authorization; and ultimately courts, where disputes arise about whether a proposed use meets the statutory criteria. The governance logic is one of distributed accountability under legal standards, rather than concentrated administrative gatekeeping. Whether this is sufficient, specifically whether the standards are determinate enough to be applied consistently, and whether the verification mechanisms are robust enough to prevent circumvention, remains to be seen, and will depend significantly on how the reform’s implementing provisions are drafted and on the judicial culture that develops around them.